Beruflich Dokumente
Kultur Dokumente
27-36
The Micro-Slip Damper Stiffness Effect on the Steady-State Characteristics of Turbine Blade
Moneer H. Tolephih , Assistant Pro.
and mechanical properties of the blade-damper system, the dry friction between mating surfaces of the damper play a role in describing the response behavior of the system[3]. The problem of friction modeling for reliable analysis of the process, becomes the major agency in relevant scopes of such subjects. The macroslip modeling type by Bazan et al [4], Ferri and Dowell[5], Muszynska and Jones[6], Wang et al [7], and the Microslip modeling by Meng et al [8,9,10], nonlinear vibration analysis for one dimensional dynamic microslip friction model and multi blade model by Cigeroglu et al [11,12] and Gabor[13], represent the professional attempts in this field. Each modeling approach bears lot of merits and some other demerits. Most advantages of the microslip friction modeling over the macro one, reflected in sophisticated representation of motion-dependency of the damping itself that can not take place in the macroslip class of friction. The excellent theoretical work, in this issue, may be found in references[8,13] where a proposition of one-bar friction modeling was displayed thoroughly, while the work of Meng et al [9] and Gabor[13] succeeds in creating a developed two-bar microslip model. The main difference between the two models is limited by the number of slid regions occupied in the friction plate of the damper which contains, for both models, one stuck region. However, the result governing equation of motion seems identical in the form regardless the type of friction bar modeling. In spite of that, there exists some difficulty in estimating the equivalent damper stiffness for the two model. The two-bar model show much lengthy manipulation of stiffness compared with the one-bar model. A group of input data should be prepared correctly for such analysis, these are consisting of all necessary parameters affecting the system response (the discrete lumped
Abstract:
In this paper, a comprehensive study of friction damper stiffness effects on the response characteristics of a typical turbine blade executing steady-state motion, is explored. The damper is modeled as a one-bar microslip type assembled in the intermediate platform attachment of the blade leaving the other attachment of a shroud mass at the blade tip to be free. A discrete lumped mass approach, previously theorized in another paper, is employed to predict the response amplitudes as well as the slip length parameter at any state of the forced frequency including the resonance condition. The analysis covers a practical range of damper stiffness values adapted from relevant studies in this field. The present main outputs show that a magnificent rising of the response occurs with the increase in the stiffness, the characteristic behavior varies appreciably and the resonant amplitudes tend to increase linearly at high levels of damper stiffness, whereas the corresponding frequency and slip length show almost uniform trend. The results can serve very well for design and control purposes in the pre-manufacture stages of the given blade-damper system.
Tolephih 27
masses and associated bending stiffness, the normal pressure variation in the damper mating surfaces, the Coulomb friction coefficient, the elastic constant, length and sectional area of the modeled bar). In his paper, the present author[14] investigated the effects of normal load parameters on the steady-state response comprehensively. Time comes now to extend the analysis to the friction damper effect regarding the elastic specification and geometry of the one-bar model (the physical length and sectional area). The author found that all these items can be gathered successfully in a single parameter referring to the damper longitudinal stiffness, without loss of originality, a matter which seems very important for quality control of the vibrated system and more essentially for mechanical design of the damping device itself.
q q2 q0 A, l, E F
Figure (2) The friction one-bar micro-slip model The plate, under friction action, is simulated as onebar model, as shown in Fig.(3), with sliding part of length . The plate, in Figs.(2,3), is plotted at 900 rotation with that in Fig.(1). The piece-wise equation of motion, for the first and second mass of the given system, can be simplified in the following forms (see Ref.[15]):
2. Theoretical Analysis:
Keeping pace with the theoretical works of references[13], the present system of 2-degree of freedom, under consideration, is schematically shown in Fig.(1), with general notations used for analysis purpose. The damper is attached to the lower platform of mass, m1, while a shroud mass, m2, is kept at the free end where an exciting harmonic force, P, is acting independently. The friction damper has an equivalent stiffness, keq, and damping constant, ceq. Fig.(2) illustrates a possible normal pressure, q , acting on the mating surface of the damper plates whose Young modulus is denoted by E, length and sectional area by l and A respectively.
l F Ff x
m1 .x1 c eq .x1 m 2 .x 2 P
k1 .x1 0 k 2 .( x 2
t t t
k 2 .( x1
x 2 ) k eq .x1 0
x1 ) P
x1
Pa .e j
x2
x1 x2
x1a .e j x 2 a .e j
k1 ceq
m1
k2 keq
m2
where x1 and x2 represent the displacement functions for the lower and upper attachments respectively, k1 and k2 the lumped bending stiffness respectively, Pa, x1a and x2a are the amplitudes of the exciting force and the piece-wise displacements respectively, while is the external frequency of the applied force and that j is the usual imaginary root and t is the elapsed time of excitation. To solve for the main response parameters x1a, x2a, then the substitution of the third of eq.(1), back into the first two equations, yields :
Turbine Blade
28
x 1a x 2a C1 C2
2
2
equating u0 with x1a, as the matter should be recognized naturally. Reference [14] displays a target function , very useful to determine upon usage of last idea, in the form of:
1 Q2 4 EA/ l 2 Q0 3
2 2
) c eq . j m1 . )
k eq
Q0
Among all input data of (Pa, k1, k2, m1, m2 and ) the indirect data keq and ceq, appearing in eq.(2) were usually estimated using Lazan[16] formula in the form of:
2
) k2 2
keq
F0 u0
ceq
3
The introduced quantities Q0 and Q2 are just alternative forms of q0 and q2 respectively in nondimensional fashion (with m denotes friction coefficient). The employment of eq.(5) needs further numerical method to assign the true value of . An iteration procedure of extended bi-secant technique, familiarly found in related fields, may be very active to achieve the goal. At the end, the present computed results of x1a and x2a as varied with , would be conveniently altered to nondimensional quantities of 1, 2 and respectively in the form of:
where F0 and u0 are the amplitudes of the translated force and corresponding displacement, at the damper bar tip respectively (refer to Fig.(3)). In reference [14], a thorough derivation of both baranthes terms, in eq.(3), has been achieved successfully. The final expressions of these items may be summarized below:
F0 u0
( EA / l )
1 1 2
2 q2 3 3 q0 q2 4 q0 3
2
2
,
14 2 5
2 2
q2 1 4 q0 1 2 q2 q0 1 2 l q2 4 q0 3
2
e1
x1a , x1s k2 m2 k2 m2 2
2
x2a x2 s
1/ 2
k1
4
e1
k2 m1 2
ceq
2( EA / l ) 3
8 7
16 2 5
2 2
k1
k2 m1
4 k1 k 2 m1 m2
q2 4 q0 3
x1s x2 s
Pa k1 k eq Pa k2 x1s
Noting that the quantity (q2/q0) is simply the normal load ratio and is the slip length ratio. The damper plate mechanical properties are evidently declared by the term (EA/l) appearing in above equation. It can be referred to one parameter entitled as the damper longitudinal stiffness whose effect stands as the main objective of the current work. In order to utilize eq.(4) the value of must be preestimated. Gabor[13] has solved this problem by NUCEJ, Vol.10, NO.1
where e1 denotes the first eign-frequency of the free-damped system, while x1s and x2s represent the static amplitudes respectively.
eight selected values of Q0=5.6, 16,32,80,160,320,800 and 8000 with Q2/Q0=-0.5. These are the actual constant parameters held well by [13,15]. In these references, the damper plate properties are kept constant with EA=40000N and l=0.2m. In the present computation this is identical to EA/l=200000. Therefore, a choice of nine distinct values of (12500, 25000, 50000, 100000, 200000, 500000, 800000 and 860000), for this parameter, seem adequate to estimate the entire effect on the system. The plan is then devoted to spread the huge out printings of 1, 2 and as related with for given (EA/l) value, as well as the corresponding resonant parameters 1res, 2res, res and res as varied with the same damper stiffness. Figs.(4-15) and Tables(1-4) satisfy this condition briefly. In Figs.(4,5) the variation of 1 with is plotted for fixed Q2/Q0 but for all the different values of (EA/l). Fig.(4) takes the lowest set value of Q0=5.6 whilst Fig.(5) takes the largest one Q0=8000. As seen, the behavior is altered obviously. The peak points (resonant) go down with (EA/l) when Q0 is small, whereas they go up for large Q0. The same trend can be noticed for 2 parameter as shown in Figs.(6,7) respectively and also for parameter in Figs.(8,9) where at approaches unity (from left or right), the peak tends to equal one (i.e. the damper plate would totally slide). Tables(1-4) show a collection of resonant values of all the main parameters for a variety of settings of (AE/l) and Q0 values keeping Q2/Q0=-0.5 as mentioned before. In each table, the data in the fifth row correspond to those recomputed from Gabor[13]. The figures, afterwards, relate to the resonant values as functions of the damper stiffness. The first two ones, in Figs.(10,11), show 1res variation for two setting values of Q0 respectively. As shown, Fig.(10) illustrates a decreasing trend of 1res with (AE/l) for Q0=5.6 to 80, whereas Fig.(11) shows counter-wise linear trend for larger Q0 value. The characteristic behavior, in this manner, seems new. The resonant curve decreases with Q0 in Fig.(10), but it increases appreciably in Fig.(11). A similar style is observed in Figs.(12,13) for 2res parameter respectively with one exception that the counter-wise increase in 2res does not seem to be perfect linear. The last Figs.(14,15) show the variation of res and res with (EA/l) respectively for all the nine set values of Q0. In Fig.(14) the slip length ratio varies linearly with very slight increase, besides the curve itself goes down with Q0. A reversed behavior is noticed in Fig.(15) for res variation, where the ratio shows large increase, while it goes up as Q0 increases further.
(a) The 1 and 2 response curves show special shifting with the increase in the exciting frequency as the damper stiffness increases. The artificial shift, for 1, is in the down-right direction (i.e. the peaks decrease and move right with ), whereas for 2 the shift is in the up-right direction. (b) The response curve rises with the damper stiffness and shows peak values near =1 when Q0 is small. The stuck part length approaches maximum value at this situation. For large value of Q0 the response curve goes up to the right with the increase in . (c) The resonant displacements 1res and 2res show two different trends with (AE/l). The first one concerns comparable small Q0 value, where the curve goes down with both Q0 and (AE/l). The second one shows counter-wise manner absolutely with linear increase as (AE/l) increases. (d) The resonant curves, of slip length ratio res and frequency ratio res, give a humble linear rising with (AE/l), but is strongly affected by Q0 value. The characteristic curve of res goes down with the increase in Q0, whilst res curve shows a trend opposite to that comparably.
5. References:
1. Srinivasan A.V. and Cutts D.G., Dry Friction Damping Mechanisms in Engine Blades , Trans. ASME, J. Engg. Power, 1983, Vol. 105, pp. 332341. 2. Menq C.H., A Review of Friction Damping of Turbine Blade Vibration , Int. J. Turbo & Jet Engines, 1990, Vol. 7, pp. 297-307. 3. Cameron T.M., An Alternating Frequency/Time Domain Method for Calculating The SteadyState Response of Nonlinear Dynamic Systems , Trans. ASME, J. Appl. Mech., 1989, Vol. 56, pp. 149-154. 4. Bazan E., Beilak J. and Griffin J.H., An Efficient Method for Predicting The Vibratory Response of Linear Structures With Friction Interfaces , Trans. ASME, J. Engg. Gas Turb. & Power, 1986, Vol. 108, pp. 633-640 5. Ferri A.A and Dowell E.H., Frequency Domain Solutions To Multi-Degree-of-Freedom, Dry Friction Damped Systems , J. Sound & Vibr., 1988, Vol. 124, pp. 207-224. 6. Muszynska A. and Jones D.I., On Tuned Bladed Disk Dynamics: Some Aspects of Friction Related Mistuning , J. Sound & Vibr., 1983, Vol. 86, pp. 107-128. 7. Wang J.H. and Sheih W.L., The Influence of a Variable Function Coefficient On The Dynamic Behavior of a Blade With a Friction Damper , J. Sound & Vibr., 1991, Vol. 149, pp. 137-145.
4. Conclusions:
In brief, the pertinent remarks, listed below, represent the main conclusion drawn from present analysis and discussion:
Turbine Blade
30
8. Menq C.H., Griffin J.H. and Beilak J., The Forced Response of Shrouded Fan Stages , Trans. ASME, J. Vibr. Acous. Stress & Reliab. In Design, 1986, Vol. 198, pp. 50-55. 9. Menq C.H., Bielak J. and Griffin J.H., The influence of Microslip On Vibratory Response, Part I: A New Microslip Model , J. Sound & Vibr., 1986, Vol. 107, pp. 279-293. 10. Menq C.H., Bielak J. and Griffin J.H., The influence of Microslip On Vibratory Response, Part II: A Comparison With Experimental Results , J. Sound & Vibr., 1986, Vol. 107, pp. 295-307. 11. Cigeroglu E., Lu W.and Menq C.H., OneDimensional Dynamic Microslip Friction Model , J. Sound and Vibr.,2006,Vol 292,pp881 898 12. Cigeroglu E, and Ozguven H., Nonlinear Vibration Analysis of Bladed Disks with Dry
Friction Dampers , J. Sound and Vibr.,2006,Vol 295, pp 1028 1043 13. Csaba G., Modelling Microslip Damping And Its Influence On Turbine Blade Vibrations, Dissertation No. 519, Linkoping Studies in Science and Technology, Linkoping University, Sweden, 1998. 14. Tolephih M.H., The Normal Load Effect on Response Characteristics of Turbine Blades with One-Bar Microslip Friction Damper , sent for publication. 15. Timoshenko S.P, Young D.H. and Weaver W., Vibration Problems In Engineering, 4th Edition, 1974, John Wiley & Sons Inc., Canada. 16. Lazan B.J., Damping of Materials And Members in Structural Mechanics, 1968, Pergamon Press Inc., London.
Figure (4) Variation of resonant (Phi1)with the normalized frequency (Omega) for different damper stiffness values (EA/L)and Fixed normalized loading of Q0 & Q2/Q0.
Tolephih 31
Figure (5) Variation of (Phi1) with the normalized exciting frequency (Omega) for different damper stiffness values (EA/L) and fixed normalized loading of Q0 & Q2/Q0.
Fig.(6) Variation of resonant (Phi2) with the normalized exciting frequency (Omega) for different damper stiffness values (EA/L) and fixed normalized loading of Q0 & Q2/Q0.
Figure (7) Variation of (Phi2) with the normalized exciting frequency (Omega) for different damper stiffness values (EA/L) and fixed normalized loading of Q0 & Q2/Q0.
Turbine Blade
32
Figure (8) The resonant (Delta) as varied with normalized exciting frequency (Omega) for different damper stiffness values (EA/L) and fixed normalized loading Q0 & Q2/Q0.
Figure.(9) The normalized (Delta) as varied with normalized exciting frequency (Omega) for different damper stiffness values (EA/L) and fixed normalized loading Q0 & Q2/Q0.
Figure (10) Variation of resonant (Phi1) with damper stiffness (EA/L) for fixed normalized load (Q2/Q0) and the first four values of the normalized load ((Q0).
Tolephih 33
Figure (11) Variation of resonant (Phi1) with damper stiffness (EA/L) for fixed normalized load (Q2/Q0) and the second four values of the normalized load (Q0).
Figure (12) Variation of resonant (Phi2) with damper stiffness (EA/L) for fixed normalized load (Q2/Q0) and the first six values of the normalized load (Q0).
Figure (13) Variation of resonant (Phi2) with damper stiffness (EA/L) for fixed normalized load (Q2/Q0) and the last two values of the normalized load (Q0).
Turbine Blade
34
Table(1). The normalized amplitudes, slip lengths and frequencies at resonance for variety of damper stiffness values (EA/L) and two normal load coefficients (Q0=5.6, 16).
Curve
1 2 3 4 5(Gabor) 6 7 8 9
EA/L
1,res
Q0=5.6
2,res a,res res 1,res 2,res
Q0=16
a,res res
Table(2). The normalized amplitudes, slip lengths and frequencies at resonance for variety of damper stiffness values (EA/L) and two normal load coefficients (Q0=32, 80).
Curve
1 2 3 4 5(Gabor) 6 7 8 9
EA/L
1,res 2,res
Q0=32
a,res res 1,res
Q0=80
2,res a,res res
Table(3). The normalized amplitudes, slip lengths and frequencies at resonance for variety of damper stiffness values (EA/L) and two normal load coefficients (Q0=160, 320).
Curve
1 2 3 4 5(Gabor) 6 7 8 9
EA/L
1,res
Q0=160
2,res a,res res 1,res
Q0=320
2,res a,res res
Table(4). The normalized amplitudes, slip lengths and frequencies at resonance for variety of damper stiffness values (EA/L) and two normal load coefficients (Q0=800, 8000).
Curve
1 2 3 4 5(Gabor) 6 7 8 9
EA/L
1,res
Q0=800
2,res a,res res 1,res
Q0=8000
2,res a,res res
Tolephih 35
Turbine Blade
36
This document was created with Win2PDF available at http://www.daneprairie.com. The unregistered version of Win2PDF is for evaluation or non-commercial use only.